The invention relates to a global vehicle track detection method based on multi-source data fusion, and belongs to the technical field of traffic detection. According to the method, three-dimensional coordinate information obtained by a millimeter wave radar and two-dimensional coordinate information obtained by a visual sensor are fused and mapped and converted to the same space coordinate system, time axis alignment is carried out based on standard time, a time-space synchronization data frame is generated, and an Euclidean distance is calculated to generate a target incidence matrix. Determining a coordinate matching relationship through the matrix, and fusing the data to generate a target track detection data set; and training a vehicle track detection model by using the data set, and carrying out track point detection by using the trained model. According to the invention, the advantages of the two sensors are complementary, the problems of missing detection and false detection of a single sensor in a complex traffic scene are solved, the consistency of data of different sensors in space is ensured, the space precision of vehicle track point detection is improved, and high-precision and multi-scene adaptive vehicle track detection can be realized.
本发明涉及一种多源数据融合的全域车辆轨迹检测方法,属于交通检测技术领域。本发明通过融合例如毫米波雷达获得的三维坐标信息和例如视觉传感器获得的二维坐标信息,将二者映射和转换至同一空间坐标系,并基于标准时间进行时间轴对齐,生成时空同步数据帧,计算欧氏距离生成目标关联矩阵,通过矩阵确定坐标匹配关系,并融合数据生成目标轨迹检测数据集;利用该数据集训练车辆轨迹检测模型,利用训练好的模型进行轨迹点检测。本发明将两种传感器的优势互补,解决单一传感器在复杂交通场景中漏检和误检的问题,确保不同传感器数据在空间上的一致性,提高车辆轨迹点检测的空间精度,可实现高精度、多场景适应的车辆轨迹检测。
Universe vehicle track detection method based on multi-source data fusion
一种多源数据融合的全域车辆轨迹检测方法
2025-05-16
Patent
Electronic Resource
Chinese
IPC: | G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G01S RADIO DIRECTION-FINDING , Funkpeilung / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung |
Multi-source data fusion track state detection method and device based on deep learning
European Patent Office | 2021
|Multi-source aircraft track data fusion method and device
European Patent Office | 2025
|Vehicle track detection system and method based on multi-sensor fusion
European Patent Office | 2024
|Universe vehicle track splicing method based on double matching
European Patent Office | 2023
|Vehicle state data acquisition method based on multi-source data fusion
European Patent Office | 2025
|